Assessment of rhizosphere processes for removing water-borne macrolide antibiotics in constructed wetlands
Limited information is available on plant rhizosphere processes for removing antibiotics in antibiotic-contaminated waters. This study identifies rhizosphere processes and evaluates their relative contributions for the macrolides (ML) removal in aquatic plant systems.
A flask-scale experiment (100 and 300 μg/L ML) incorporating Juncus effuses and Canna indica was used to identify the root adsorption, rhizobacterial influences, and plant uptake responsible for the ML (i.e., anhydroerythromycin A, roxithromycin, clarithromycin and tilmicosin) removal.
Total ML removal rates due to rhizosphere processes were respectively 43.7–67.6% and 44.3–82.2% at 100 and 300 μg/L ML. J. effuses removed ML more effectively than C. indica (P < 0.05). The relative contribution of rhizospheric pathways to remove all ML followed the order: root sorption > rhizobacterial influence > plant uptake (P < 0.01). Sorption and rhizobacterial activity were important removal pathways in wetland plant microcosms, accounting for 36.5–72.8% and 20.5–54.2% of the total rhizosphere associated removal of ML, respectively.
Root sorption and rhizobacterial influence were the main rhizospheric pathways of ML removal in aquatic plant systems. Fe plaque on the root surface, rhizobacterial number and bacterial activity play significant roles in the removal of target pollutants.
KeywordsMacrolides Aquatic plant system Rhizospheric pathway Root sorption Rhizobacterial activity
Macrolides (ML) are important antibiotics that have been widely used in human and veterinary medicine (Feitosa-Felizzola et al. 2009; Weng et al. 2012). They are poorly metabolized by human and animal digestive systems and are therefore either excreted unchanged or transformed via the urine or feces (Yang et al. 2011; Dan et al. 2013). However, conventional wastewater treatment plants only partially eliminate these chemicals; therefore, as emerging organic pollutants, ML are frequently detected worldwide in the final effluents of Wastewater Treatment Plants (WWTPs) (Gros et al. 2007; Miao et al. 2004), in surface waters (Managaki et al. 2007; Xu et al. 2007), and even in drinking water (Ye et al. 2007). Concerns about environmental ML residues arise from their potential environmental hazards related to their continuous introduction as “pseudopersistent compounds” in the environment (Hernando et al. 2006), and their possible contribution to the selection of antibiotic-resistant bacterial strains (Silva-Costa et al. 2012; LaPara et al. 2011; Su et al. 2012; Tao et al. 2010). To reduce negative impacts on the environment and human health, it is necessary to provide more efficient control methods for these antibiotics to reduce their negative impacts on the environment and human health.
Constructed wetlands (CWs) can efficiently purify waters because they have the potential to reclaim wastewater via the removal of trace concentrations of emerging organic pollutants (Matamoros et al. 2008; Matamoros and Bayona 2006; Hijosa-Valsero et al. 2011; Dan et al. 2013). Meanwhile, wetland plants improve the removal efficiency of certain organic pollutants (Dan et al. 2013; Mei et al. 2014). Erythromycin was only removed by a planted horizontal subsurface flow wetland system (Hijosa-Valsero et al. 2011). This removal could be due to sorption, plant uptake, and partial or complete degradation processes. Tetracycline (TC) and oxytetracycline (OTC) can be completely removed from aqueous media by submerged plants such as parrot feather and floating plants such as water lettuce (Gujarathi et al. 2005). Vetiver grass is also highly effective for removing organic contaminants such as TC from hydroponic media (Makris et al. 2007). Tetracycline was detected in root and shoot tissues of vetiver grass, confirming the existence of uptake and root to shoot translocation (Datta et al. 2013). The total antibiotic content in Phragmites australis follows the sequence ciprofloxacin > oxytetracycline > sulfamethazine, and the distribution of all antibiotics follows the sequence root > leaf > stem (Liu et al. 2013). Wetland plants adsorb and absorb nutrients from wastewater and substrates, and provide large surfaces for microbial colonization, as well as enhancing rhizophysical-chemical processes for the removal of contaminants in constructed wetland treatment systems (Li et al. 2013; Matamoros et al. 2012). Plants play an important role in enhancing desirable microbial functional groups in CWs (Gagnon et al. 2012; Faulwetter et al. 2009), resulting in the removal of pharmaceutical pollutants (Zhao et al. 2015). However, Weber et al. (2011) found that ciprofloxacin had an adverse effect on the inherent bacterial communities in wetland systems, resulting in the reduced ability to assimilate anthropogenic carbon-based compounds.
Wetland treatment technologies rely on the interactions between chemical, physical, and biological processes to reduce pollutant concentrations. Compared with the simplistic and “insignificant” plant effects on organic pollutants found in previous studies (Tang et al. 2015; Dan et al. 2013), quantification of rhizosphere-associated processes, such as those investigated in this study, has been limited (Reinhold et al. 2010; Chen et al. 2015). It is difficult to draw general conclusions about possible plant uptake, root sorption of antibiotics, and rhizobacterial interactions with these pollutants because the data available in the literature for distinguishing and quantifying different plant rhizosphere associated removal pathways are limited. Consequently, in this study, a flask-scale experiment (100 and 300 μg/L ML) of two wetland plant species, Canna indica and Juncus effuses (Mei et al. 2014; Li et al. 2013; Konnerup and Brix 2010), was used to identify and quantify the rhizosphere-associated processes responsible for the aqueous depletion of select antibacterial agents: anhydroerythromycin A (ETM-H2O), roxithromycin (RTM), clarithromycin (CTM), and tilmicosin (TIL). The study was performed under hydroponic conditions to avoid the potential interference of soil or other wetland substrate particles that could adsorb the antibiotics (Liu et al. 2013). This is the conventional culture method for such studies (Blamey et al. 2014; Reinhold et al. 2010). Additionally, the study focused on distinguishing between living and dead plant processes and on exploring the potential factors that contribute to the aqueous depletion of ML in aquatic plant systems.
Materials and methods
Collection and maintenance of wetland plants
Juncus effuses is a typical emergent wetland plant. Ornamental flowers such as Canna spp. and other species (Brix et al. 2007; Konnerup and Brix 2010; Zurita et al. 2009) are used in CWs to increase the aesthetics of the system. Canna spp. can grow in both terrestrial and aquatic environments (Konnerup and Brix 2010). These plant species can be used for the treatment of wastewater in constructed wetlands and are known to be highly tolerant to various types of wastewater (Kumwimba et al. 2016). However, little is known about their preference and response to the antibiotic pollution in water. Two plant species (approximately three weeks old) were collected from a nursery and propagated in 50% Hoagland’s nutrient solution: 2.5 mM Ca (NO3)2·4H2O, 2.5 mM KNO3, 1.0 mM MgSO4·7H2O, 0.5 mM KH2PO4, 25 μM Fe (II)-EDTA, 22.5 μM H3BO4, 0.5 μM ZnSO4·7H2O, 0.25 μM CuSO4·5H2O, 5.0 μM MnSO4·H2O, 0.05 μM (NH4)6Mo7O24·4H2O (Hoagland and Arnon 1950). All plants were acclimated to the greenhouse conditions at least 30 days before the experiment. Individuals of the same species with similar shoot height and weights (approximately 20 cm in height and 10–15 g in wet weight) were selected for a hydroponic experiment. Each treatment, each individual plant was set in triplicate and arranged randomly in a greenhouse environment at actual room temperature (35/30 ± 2 °C day/night) and a 12 h photoperiod per day.
Chemicals and hydroponic test system set-up
The target compounds were selected predominantly based on their usage in China. They included anhydroerythromycin A, roxithromycin, clarithromycin, and tilmicosin. Supplementary Table S1 summarizes the physiochemical properties of the target antibiotics. Target standards (anhydroerythromycin A, roxithromycin, clarithromycin, and tilmicosin) and an internal standard (josamycin) were purchased from Dr. Ehrenstorfer GmbH (Germany). HPLC grade methanol was obtained from Merck (Germany). Stock solutions of antibiotics were dissolved at 100 mg/L in methanol and stored at 4 °C. The working standard solutions were prepared from this solution at 1.0 mg/L.
Exposure concentrations of 100 and 300 μg/L were chosen to avoid inhibition of test plants and allow for the identification of root-associated removal pathway. Sorption and uptake could be concealed at lower concentrations during the short experimental period (Reinhold et al. 2010; Liu et al. 2013; Chehrenegar et al. 2016). As in similar studies (Reinhold et al. 2010; Tront and Saunders 2006; Chehrenegar et al. 2016), 18 plants of a similar size were selected from each species and subjected to four treatments: (i) a control with 100 and 300 μg/L of ML, containing no plant and no bacteria, to test the physicochemical removal of pollutants via hydrolysis (n = 3); (ii) a living plant treatment with 100 and 300 μg/L of ML, with one healthy plant to measure the total removal processes in the plant system (n = 3); (iii) a dead plant treatment with 100 and 300 μg/L of ML, with one chemically inactivated plant exposed to 1 g/L sodium azide for 5–7 days and rinsed prior to addition in the experimental treatments to measure the root sorption in a wetland plant system (n = 3); and (iv) a rhizobacterial treatment with 100 and 300 μg/L of ML, with the immediate addition of a microbial solution extracted from an individual or cluster of plant roots using an 0.85% NaCl solution to monitor the removal influence of ML by the epiphytic bacteria on the root surface (n = 3). All batch experiments were conducted in 1 L brown glass bottles for hydroponic treatment and filled with 0.5 L of the culture solution at pH 6 prepared using 50% Hoagland’s sterile solution.
Aqueous samples of 2 ml were taken from each treatment at 0, 0.5, 1, 2, 4, 6, 8, and 10 d during the experiment. In this study, we regularly supplemented the liquid level with sterilized water before collecting water samples to counterbalance drawdown caused by evaporation and transpiration. These water samples were refrigerated and transported to the laboratory as soon as possible, where they were filtered through a 0.45 μm membrane filter into a 2-ml amber glass vial before freeze-drying (within 48 h). The extracts were re-dissolved within 1 ml methanol and stored at −18 °C for analysis by HPLC–MS/MS.
DCB extraction of Fe plaque
Iron plaque on the fresh root surface of wetland plants in dead plant treatments before the experiment was extracted using the dithionite-citrate-bicarbonate (DCB) method (Taylor and Crowder 1983). More details can be found in Mei et al. (2014) and Cheng et al. (2014). The concentration of Fe in the extracts was determined by inductively coupled plasma optical emission spectrometry (ICP-OES; Optima 2000 DV, Perkin Elmer, USA).
Aqueous dehydrogenase activity
Enzyme activity is one way to describe the general condition of the microbial populations in an environmental matrix (Margesin and Schinner 1997; Margesin et al. 2000). Changes in the activity of aqueous dehydrogenase in the rhizobacterial treatments over the entire experimental period, including C. indica and J. effuses exposed to 100 and 300 μg/L ML, was analyzed by triphenyltetrazolium chloride (TTC) reduction to triphenyl formazone (TPF), and expressed as μg of TTC transformed per ml−1 of solution h−1. Briefly, 5 ml water samples were incubated in 2 ml 1% TTC and 2 ml 0.1 M glucose at 37 °C for 24 h. The reactions were terminated by 0.25 ml 98% H2SO4, and the products were extracted with 5 ml toluene on a shaker for 30 min. After centrifugation, the TF dissolved in the toluene was assayed at 487.5 nm. The activity was measured following a modified method of Zhou et al. (2005).
Enumeration by microscopy
The total number of bacterial cells in the rhizobacterial treatments was determined by epifluorescence microscopy after staining with DAPI (4, 6-diamino-2- phenylindoldihydrochlorid-dilactate) on separate filters (black 0.22 mm polycarbonate filters, Millipore). DAPI staining was applied after fixation of the aqueous samples with paraformaldehyde by adding 1 μg/ml DAPI to a final ratio of 10% for 30 min (Nielsen and Nielsen 2002). The samples were viewed with a Leica DMR microscope (22DI-E-D282) (Supplementary Fig. 1).
Calculation of plant uptake of ML in the living plant systems
HPLC-MS/MS analytical procedure
The liquid chromatography system was an HP 1100 LC (Agilent Technologies, USA) controlled gradient system. It was equipped with an auto sampler, a pump, and a column oven with a thermostat. A ZORBAX SB-C8 (3.5 μm, 2.1 × 150 mm) chromatograph column was employed and was operated at 25 °C. Optimum separation was achieved using gradient elution. The mobile phase consisted of water containing 2‰ formic acid (A) and acetonitrile containing 2‰ formic acid (B). The gradient was set up as follows: 88% A (0 min), 65% A (1 min), 50% A (2 min), 20% A (3 min), 20% A (7 min), 88% A (7.1 min), 88% A (17 min). The injection volume was 10 μl for each sample, and the flow rate was 0.3 ml/min. Mass spectrometric measurements were performed in an AB 4000Q TRAP mass spectrometer (AB Sciex, USA) equipped with an electrospray ionization (ESI) source that operated in the positive ionization mode. The nebulizer, air curtain, and desiccator pressure were set at 379, 138 and 310 kPa, respectively. The temperature of the dry gas was 500 °C and the source voltage was set at 5500 V. The declustering potential and collision energy were optimized to achieve the highest possible sensitivity of the instrument (Supplementary Table S2).
Quality assurance/quality control
The method detection limits (MDLs) and the method quantification limits (MQLs) were determined as the minimum detectable amount of an analyte from the environmental matrix-spiked extract in the MRM mode with a signal-to-noise ratio of 3 and 10, respectively (Jelić et al. 2009; Shao et al. 2009). The MDLs of the target antibiotics in the culture solution with or without plants were 0.08–0.35 μg/L and 0.08–0.88 μg/L, respectively. The MQLs were 0.26–1.15 μg/L and 0.26–2.93 μg/L, respectively. Recovery experiments were performed by spiking the standard solutions with 50% Hoagland’s solution grown with or without plants. The recoveries of the target compounds from these environmental matrix samples ranged between 80.5% and 135.0%, and the relative standard deviation (n = 4) was less than 11.6% (Supplementary Table S3). The matrix effects were calculated by spiking the standard solutions at a concentration of 100 μg/L of each compound into the sample extracts of the seven environmental matrices and comparing the obtained concentration obtained in the matrix to the concentration in the solvent (Grujic et al. 2009). Values of greater or less than 100% indicate signal enhancement or suppression, respectively. The matrix effect was 87.1–130.4% in the solution matrices. Quality assurance/quality control (QA/QC) was ascertained from sampling to analysis. Procedural blanks, solvent blanks, and spikes, as well as QC samples, were included in the extraction and analysis.
The SPSS 13 software package (Chicago, USA) was used for statistical analysis of the data. A three-way multiple analysis of covariance (MANCOVA) at P ≤ 0.05, using the exposure concentration of the test compounds, the plant species, and the treatments as the three fixed factors, sampling time as the covariate, and the ln C values as the multivariates, was used to test the significant of the effects of concentration, plant species, or treatment on the aqueous depletion rates of the ML. If a significant differences was found at P ≤ 0.05 between treatments, a Tukey’s HSD (honest significant difference) test was calculated to determine where the difference was. The Pearson’s correlation was used to establish determine the relationships between parameters.
Evaluation of plant growth
The relative growth rate (mg day−1) of whole plants in living plant treatments
Exposure concentration of macrolides
1.68 ± 0.81a
1.73 ± 0.62a
1.44 ± 0.75b
1.45 ± 0.27b
Microbial growth and aqueous dehydrogenase activity
Dynamics of ML in solution
ML removal pathways
The macrolide (ML) removal rates (%) for different pathways in each plant system with exposure concentrations of 100 and 300 μg/L at the end of the 10-day experiment (means ± SD, n = 3)
ML removal rates (%) for root pathways in system
Exposure concentration: 100 μg/L
40.7a ± 0.7
44.6a ± 3.0
2.5b ± 1.2
33.9a ± 4.5
2.7b ± 1.2
23.4b ± 4.1
32.3a ± 3.8
2.1bc ± 1.0
34.7a ± 9.1
29.5a ± 1.3
1.8c ± 0.9
26.5b ± 5.3
41.4a ± 11.1
3.3ab ± 1.3
16.7b ± 8.9
32.5a ± 5.5
4.1a ± 1.7
18.1b ± 3.9
12.2c ± 6.0
3.1ab ± 1.3
22.3b ± 7.0
43.2a ± 0.5
2.6b ± 1.1
Exposure concentration: 300 μg/L
11.8c ± 0.5
27.5a ± 1.5
2.3 cd ± 0.4
12b ± 4.0
24.1a ± 2.2
2.5 cd ± 0.4
13.5c ± 1.7
32.7a ± 2.3
1.7d ± 0.3
13b ± 4.7
38.3a ± 4.5
1.3d ± 0.2
7.3bc ± 4.3
21.3a ± 3.1
6.7ab ± 2.6
7.4c ± 2.9
21.2a ± 1.5
7.4a ± 2.8
10.7bc ± 2.0
28.1a ± 0.9
5.2b ± 2.1
8.5c ± 2.7
28.0a ± 5.2
4.3bc ± 1.7
The relative contribution of the different rhizospheric pathways to the removal followed the same order for all ML, that is, root sorption > rhizobacterial effect > plant uptake (Supplementary Fig. 4). Sorption was the dominant pathway for all ML removal in the plant microcosms, accounting for 36.5–72.8% of the total removal due to rhizosphere effects, with the exception of clarithromycin and tilmicosin at 100 μg/L. The rhizobacterial effect was another important removal pathway in the wetland plant microcosms, which accounted for 20.5–54.2% of the total root-effect removal of ML.
Three-way MANOVA analysis results showing the effects of exposure concentration, plant species, and treatment on the removal rates (%) of four ML, with time as the covariate
Exposure concentration * Plant species
Exposure concentration *Treatment
Plant specie * Treatment
Exposure concentration * Plant species * Treatment
In the present study, the dehydrogenase activity showed a significant positive correlation with the Log (DAPI bacterial cells) in the wetland plant rhizobacterial treatments exposed to 100 and 300 μg/L of pollution (P < 0.01) (Supplementary Fig. 6). These results indicate that dehydrogenase activity in an aquatic rhizosphere environment reflects bacterial growth.
The present study was based on hydroponic systems and revealed that root sorption was the dominant wetland plant rhizospheric pathway and that it was related to the pollutant concentrations, i.e., a higher concentration of ML enabled greater root adsorption. Similar results were also reported by Feitosa-Felizzola et al. (2009), who showed that the adsorption of clarithromycin and roxithromycin occurred on the surface of three environmental subsurface sorbents (iron (III) and manganese (IV) oxy-hydroxides), which were main compound of Fe plaque on the root surface (Armstrong 1979; Cheng et al. 2014; Yang et al. 2014). Iron plaque build-up on the root surface was caused by the release of oxygen from wetland plant roots (Taylor et al. 1984; Amils et al. 2007), as a mixed compound of iron (Fe3+) oxides/hydroxides, including two-line ferrihydrite, goethite, and lepidocrocite (Chen et al. 1980; Liu et al. 2006). The formation of Fe plaque on root surfaces has been widely observed in aquatic plants (Pi et al. 2011). The adsorption probably occurred through a surface complexation mechanism and was accompanied by slow degradation of particular ML, including hydrolyzed and dealkylated products (Feitosa-Felizzola et al. 2009; Hamdan 2003). The effects and mechanisms of Fe plaque formation around the root surface on organic pollutants, such as antibiotics in an aqueous solution, require further investigation. In addition, root sorption had no significant correlation (P > 0.05) with the hydrophobicity of the test pollutants, which further suggests that ML root adsorption under the experimental conditions of this study may not be related to hydrophobic adsorption. The relationship between the Fe plaque and its sorption effect should be verified by further experiments.
The root bacterial associated with wetland plants also plays a significant role in a hydroponic system. Some studies have reported an inhibitory effect of antibiotics on the microbial community (Liu et al. 2009) via the reduction of their ability to assimilate anthropogenic carbon-based compounds (Weber et al. 2011). In this study, a high dose of pollutant resulted in a certain stimulatory effect, which may be related to the solvent (methanol) of the target pollutant. Methanol may provide a carbon source for microbial growth (Choo et al. 2016) that promoted the rapid growth of microorganisms during the experiment, and helped to reduce the concentration of the target pollutant. The effect of the rhizobacterial activity on removing trace organic pollutants may be reflected by biodegradation (Reinhold et al. 2010; Zhang et al. 2013; Chehrenegar et al. 2016) and the biological adsorption onto live and dead bacterial and/or fungal biomass (Vasiliadou et al. 2013; Hadibarata et al. 2013). However, the present study showed that the removal of marcolides by J. effuses, which had a lower rhizosphere microbial biomass and activity was significantly better than that of C. indica (P < 0.01), possibly because J. effuses had a higher growth rate than C. indica (Table 1). The influence of the plant species on the growth and activity of rhizosphere microorganisms is important.
This study shows the options for the bioremediation of environmental pollutants related to macrolide antibiotics. This calculation for the influence of rhizobacteria and root sorption is based on that these processes are simply additive. However, this is not the case for natural wetland conditions. Furthermore, because the surfaces on which these bacteria can attach during the experimental treatment and the supply of additional C sources are lacking, inoculum addition may not reflect the actual decomposition and sorption rates; hence, the efficiency of degrading bacteria is lower if the roots are removed, as root exudates and rhizosphere environment provided supports to bacterial life (Eisenhauer et al. 2017; Wu et al. 2017). So our results very likely underestimate the actual root-associated bacterial influence. Moreover, no analyses were performed to calculate the removal concentrations of ML in the plants; therefore, no clear proof of plant uptake was obtained. These results indicate that single-process studies in wetland systems may not reflect the role of aquatic plants in the fate of pollutants (Reinhold et al. 2010). Further studies are required to clarify the fate of test microcontaminants in a natural wetland environments, including the elimination and potential production of degradation.
We are also grateful to the National Natural Science Foundation of China (51509106, 51579115), the National Science Foundation for Post-doctoral Scientists of China (2015 M572410), the Natural Science Foundation of Guangdong Province (2016A030310097), and Pearl River S&T Nova Program of Guangzhou (201710010052), China for financial support.
- Hoagland DR, Arnon DI (1950) The water-culture method for growing plants without soil. University of California, Berkeley, p 39Google Scholar
- Jelić A, Petrović M, Barceló D (2009) Multi-residue method for trace level determination of pharmaceuticals in solid samples using pressurized liquid extraction followed by liquid chromatography/quadrupole-linear ion trap mass spectrometry. Talanta 80:363–371. doi:10.1016/j.talanta.2009.06.077 CrossRefPubMedGoogle Scholar
- Reichel R, Rosendahl I, Peeters E, Focks A, Groeneweg J, Bierl R, Schlichting A, Amelung W, Thiele-Bruhn S (2013) Effects of slurry from sulfadiazine- (SDZ) and difloxacin- (DIF) medicated pigs on the structural diversity of microorganisms in bulk and rhizosphere soil. Soil Biol Biochem 62:82–91. doi:10.1016/j.soilbio.2013.03.007 CrossRefGoogle Scholar
- Tang X, Wang S, Yang Y, Tao R, Dai Y, A D (2015) Removal of six phthalic acid esters (PAEs) from domestic sewage by constructed wetlands. Chem Eng J 275, 198–205. doi:10.1016/j.cej.2015.04.029
- Zhao C, Xie H, Xu J, Xu X, Zhang J, Hu Z, Liu C, Liang S, Wang Q (2015) Bacterial community variation and microbial mechanism of triclosan (TCS) removal by constructed wetlands with different types of plants. Sci Total Environ 505:633–639. doi:10.1016/j.scitotenv.2014.10.053 CrossRefPubMedGoogle Scholar